Triton Inference Server Web UI

Manage and monitor your Triton Inference Server with a web browser. You can use this project to develop and test your models on Triton Inference Server more easily. It is also useful for monitoring your models in production but it is not recommended to edit model configuration in production since it will be lost when the server restarts.

This project is built on top of Triton Inference Server HTTP API. It is not affiliated with NVIDIA.

Features

List all models from the model repository.
Easily load/unload models from the model repository.
View & edit model configuration on the fly.
Monitor models metrics.

Installation

Docker

Pull the image:

docker pull duyvd/triton-inference-server-web-ui

Run the container:

docker run -d --name triton-inference-server-web-ui \
    -p 3000:3000 \
    -e API_URL=<api-url> \
    -e API_AUTH_HEADER=<api-auth-header> \
    duyvd/triton-inference-server-web-ui

Where:

<api-url> is the URL of the HTTP server, usually on port 8000.
<api-auth-header> is optional, if your Triton Inference Server is protected by an authentication header.

For local Triton Inference Server, you can use the following command:

docker run -it --name triton-inference-server-web-ui \
    -p 3000:3000 \
    -e API_URL=http://localhost:8000 \
    duyvd/triton-inference-server-web-ui

Finally, open your browser and go to http://localhost:3000.

Build from source

Prerequisites:

Node.js >= 18

Clone the repository and install dependencies:

git clone https://github.com/duydvu/triton-inference-server-web-ui
cd triton-inference-server-web-ui
npm install

Prepare the .env file at the root of the project:

# The URL of the HTTP server, usually on port 8000
API_URL=<api-url>
# Optional, if your Triton Inference Server is protected by an authentication header
API_AUTH_HEADER=<api-auth-header>

For local Triton Inference Server, you can use the following .env file:

API_URL=http://localhost:8000

Build:

npm run build

Run the server:

npm start

License

MIT License